Most of the AUM allocated to levered short treasury ETFs is in TBT equity
There is $7.4BN in AUM allocated to levered short ETFs of which $6.4 BN is allocated to the ticker TBT (Proshares Ultrashort 20+ Yr ETF). This ETF’s underlying is the Barclays’ 20 Yr Bond Index, whose historical chart is shown below.
The next chart shows two hypothetical paths that the underlying follows from now to 6 months into the future. Which of the two paths do you think leads to a positive payoff from buying the 2x short levered ETF and holding it for the entire duration?
The answer is neither. In Path #1, the underlying returns -3.5% over the time horizon, but a 2x levered short ETF would have returned -0.34% over the same period. This does not also account for management fees and interest expenses that the investor would also incur. In Path #2, the underlying returns -1.96%, but the 2x levered short ETF would have returned -5.5% over the same time horizon. The chart below shows three lines for each path: the trajectory of the underlying, the profitability from shorting the underlying with 2x leverage and the profitability of holding a 2x levered short ETF (TBT equity).
To be fair, holding a levered ETF past 1 day does not always lead to underperformance relative to buying/shorting the underlying with leverage. During the first month of Path#2, for example, the 2x levered short ETF outperforms shorting the underlying with 2x leverage.
Understanding what drives the divergence in performance
So what are some of the factors that drive the difference in payoff between buying/shorting the underlying on your own with leverage and buying a levered ETF? The first factor is the degree of serial correlation (streakiness) or mean reversion in the underlying. A path demonstrating a higher degree of mean reversion, all else being equal, will lead to a drag in performance of the levered ETF, while a path showing more serial correlation with give the levered ETF a bump in performance relative to levering up and buying/shorting the underlying.
The second factor is the length of your holding period: the longer you hold a levered ETF the more its performance is likely to diverge from simply levering up the underlying. The video below discusses these two factors in more detail, along with simulations of different scenarios. The video talks about the case of a 2x long levered ETF, but all the principles apply for a 2x short ETF as well.
The third factor influencing the relative under/outperformance of a levered ETF is the size of the moves in the underlying. The larger the daily moves in the underlying, the more a levered ETF will be likely to diverge in performance from simply levering the underlying.
The fourth factor we discuss is the amount of leverage in the ETF. A more highly levered ETF will cause more price divergence than a less levered ETF. These last two factors are discussed in more detail and with the aid of computer simulations in the video below:
Many investors who are betting on price declines in treasuries are buying levered short ETFs and holding them past one day. When investors do this they are making a highly complex bet on the future path and volatility of the underlying bond index. Therefore if investors want to short treasuries using a levered ETF, they should think through what path they expect treasuries to follow and whether those price declines are likely to take place during a high or low volatility regime.
If the expected decline in treasuries coincides with higher volatility and/or a higher degree of mean reversion in bond prices (perhaps reflecting short term uncertainty in the market), then holding onto a levered short treasury ETF over that period will likely underperform simply shorting an unlevered bond fund (e.g TLT equity) using leverage. It is entirely possible that investors who end up being right on the direction of the underlying bond index and would have profited from shorting it, may instead lose money buying a levered short ETF.
Understanding the risks of levered ETFs, by Robert Jarrow
Leveraged ETFs and High School Level Mathematical Truths, by Kid Dynamite
Path Dependance of levered ETF returns, by Avellaneda and Zhang
Why the SEC should look at levered ETFs, by Felix Salmon (Reuters Blog)